Dear all,
I am trying to do a fair comparison of the Differential expression results that I have obtained using two different approaches, one using Sleuth pipeline and the second one using EgdeR. However, I realized that Sleuth reports beta values instead of the worldwide used log Fold changes, which complicates a bit the comparison. In Sleuth documentation, they refer to the'beta' values as the effect size, 'technically a biased estimator of the fold change'. However, I have read it is not exactly the same and they should NOT be directly compared...
I have tried to compute from scratch my own log Fold change values from the normalized counts, but I do not get exactly the same results as EdgeR, apparently, the reason why this is happening is because edgeR does not simply take the averages of individual cpm values (http://seqanswers.com/forums/showthread.php?t=26610), it is also taken into account other parameters such as the library size or the size of the counts...
Any ideas or suggestion to solve it?
Thank you in advance,
Cristina
